Antimicrobial resistance and documented empirical antibiotic prescribing among cancer patients with culture-confirmed infections at a tertiary referral hospital in Rwanda, 2019–2024

Abstract Background Antimicrobial resistance (AMR) threatens infection management in cancer patients, yet data from sub-Saharan African oncology settings are scarce. We described the microbiology, resistance patterns and documented empirical prescribing among cancer patients with culture-confirmed infection in Rwanda. Methods We conducted a retrospective, records-based study of 297 culture-confirmed infection episodes in 286 cancer patients at King Faisal Hospital, Kigali, from 2019 to 2024. Routine susceptibility results were interpreted with the Clinical and Laboratory Standards Institute breakpoints in use at the time. Multidrug resistance (MDR) was classified with the organism-specific definitions of Magiorakos et al., which covered 232 of 254 bacterial isolates. Factors associated with MDR were examined by modified Poisson regression (adjusted prevalence ratios, aPR) with sensitivity analyses. Empirical regimens documented in the electronic medical record (EMR; 105 episodes, 100 inpatient) were compared with the WHO AWaRe framework and the 2022 Rwanda Standard Treatment Guidelines. Results Urinary tract infections (UTIs) predominated (172/297; 57.9%). Escherichia coli (101/297; 34.0%), Klebsiella spp. (32/297; 10.8%) and Enterococcus spp. (25/297; 8.4%) were the commonest isolates, and ESKAPEE pathogens accounted for 201/297 episodes (67.7%). MDR was identified in 138 of 232 covered isolates (59.5%; 95% CI 53.1–65.6) and 138 of 214 classifiable isolates (64.5%; 95% CI 57.9–70.6); restricting to more complete panels increased the estimate. Ceftriaxone resistance was high ( E. coli 65/87, 74.7%; Klebsiella spp. 20/29, 69.0%), and carbapenem resistance in Acinetobacter spp. was 8/13 (61.5%). MDR prevalence did not differ by infection type (aPR versus UTI: bloodstream 0.94, 95% CI 0.67–1.31; skin and soft tissue 0.98, 95% CI 0.72–1.34) and was lower among gram-positive isolates (aPR 0.42, 95% CI 0.21–0.85), owing to Enterococcus spp. Adherence of documented regimens to the national guideline was lowest for UTIs (1/43; 2.3%). Outpatient empirical prescribing was rarely documented in the EMR and could not be evaluated. Conclusions MDR was common across all infection types, and ceftriaxone, the most frequently documented empirical agent, showed limited in vitro activity against the prevailing isolates. Whether oncology-specific empirical protocols based on local antibiograms are warranted should be evaluated prospectively, including clinical outcomes, alongside AWaRe-aligned stewardship and better documentation of prescribing.